Thematic CERN School of Computing on Scientific Software for Heterogeneous Architectures

5 Oct 2025, 15:00 → 11 Oct 2025, 11:30 Europe/Zagreb

Medils Institute, Split

Alberto Pace (CERN), Kristina Gunne (CERN), Andrzej Nowicki (CERN)

The 18^th Thematic CERN School of Computing (tCSC 2025) will take place on October 5-11, 2025. The theme of the school is "Scientific Software for Heterogeneous Architectures" - see the academic programme for more details.

This school is organized by CERN in collaboration with the Faculty of Science, University of Split. The school will take place in Split, Croatia, and be hosted at the Mediterranean Institute for Life Sciences (MEDILS) Conference Centre. The Centre is a historical renovated building situated in a wooded and landscaped park located on the Adriatic Sea coast, a few kilometers from the center of Split.

Target Audience

The School is aimed at postgraduate (i.e. a minimum of Bachelor's degree or equivalent) students, engineers, and scientists with a few years of experience in particle physics, in computing, or in related fields. We welcome applications from all countries and nationalities.

Important dates

• 14 May application opens
• 10 July application close
• 23 July invitations sent to selected students
• 20 August confirmation of participation deadline
• 3 September participation fee deadline  

 

tCSC in Split - Poster.pdf

Sunday 5 October

15:00 → 16:00 Registration

16:00 → 16:20 Welcome and introduction

16:20 → 17:00 Self presentations

One minute-one slide self presentations

17:00 → 17:30 Transport to Split

17:30 → 18:45 Guided tour of Split

19:30 → 21:00 Welcome dinner at Restoran Para di šoto

Monday 6 October

07:45 → 08:45 Breakfast

08:45 → 09:45 Opening Session

Speaker: Alberto Pace (CERN)

09:45 → 10:45 Preparing for the HL-LHC computational challenge

In this talk we will introduce some basic concepts related to HEP data processing and analysis workflows, seeing them in action in the context of LHC experiments. We’ll also talk about the evolution of the LHC accelerator and experiments. We’ll characterise at a high level what are the consequences of those upgrades for the HEP data processing software, in particular in the context of an evolving hardware and computing infrastructure.

Speaker: Danilo Piparo (CERN)

10:45 → 11:00 Announcements

11:00 → 11:30 Break

11:30 → 12:30 CPU Hardware Architecture and Evolution

CPU Architecture and High Performance - lecture 1

• Hardware evolution of the CPU
• Memory hierarchy, caching, NUMA
• Microarchitecture of modern CPUs

Speaker: Guilherme Amadio (CERN)

12:30 → 13:30 Performance Analysis on Modern CPUs

CPU Architecture and High Performance - lecture 2

• Performance analysis tools for Linux
• CPU features for performance analysis
• Top-down microarchitecture analysis

Speaker: Guilherme Amadio (CERN)

13:30 → 14:30 Lunch

14:30 → 16:00 Study or Sports time

16:00 → 16:30 Break

16:30 → 17:30 CPU Architecture and High Performance - exercise 1

Speaker: Guilherme Amadio (CERN)

17:30 → 18:15 Student lightning talks

19:30 → 20:30 Dinner

Francuska 7, Belgrade 11000, Serbia

Tuesday 7 October

07:45 → 08:45 Breakfast

08:45 → 09:45 Writing parallel software

Parallel and Optimised Scientific Software - lecture 1

• Amdahl's and Gustafson's laws
• Asynchronous execution
• Finding concurrency, task vs. data parallelism
• Using threading in C++ and Python, comparison with multi-process
• Resource protection and thread safety
• Locks, thread local storage, atomic operations

Speaker: Sebastien Ponce (CERN)

09:45 → 10:45 Writing efficient software

Parallel and Optimised Scientific Software - lecture 2

• Why Python and C++ ?
• Recent evolutions: C++ 11/14/17
• Modern features of C++ related to performance
• Templating versus inheritance, pros and cons of virtual inheritance
• Python 3, and switching from Python 2

Speaker: Sebastien Ponce (CERN)

10:45 → 11:00 Announcements

11:00 → 11:30 Coffee break

11:30 → 12:30 Low-level Performance Optimization Guidelines

CPU Architecture and High Performance - lecture 3

• Main sources of performance bottlenecks
• Floating point arithmetics performance
• Advanced low-level performance tuning

Speaker: Guilherme Amadio (CERN)

12:30 → 13:30 CPU Architecture and High Performance - exercise 2

Speaker: Guilherme Amadio (CERN)

13:30 → 14:30 Lunch

14:30 → 16:00 Study or sports time

16:00 → 16:30 Coffee break

16:30 → 17:30 Optimizing existing large codebase

Parallel and Optimised Scientific Software - lecture 3

• Measuring performance, tools and key indicators
• Improving memory handling
• The nightmare of thread safety
• Code modernization and low level optimizations
• Data structures for efficient computation in modern C++

Speaker: Sebastien Ponce (CERN)

17:30 → 18:30 Data-Oriented Design

CPU Architecture and High Performance - lecture 4
- Principles of data-oriented design
- Memory access and data-type profiling
- Data structure performance optimization

Speaker: Guilherme Amadio (CERN)

18:30 → 19:30 Managing Large Databases: Challenges and Lessons

With around 100 Oracle databases and several PB of data, CERN faces numerous intriguing challenges and unique caveats in regards to managing the database infrastructure. This lecture aims to provide an overview of both past and current issues and obstacles encountered in the realm of database systems at CERN. By doing so, it seeks to offer valuable insights into potential problems to avoid and effective solutions to address encountered difficulties.

Speaker: Andrzej Nowicki (CERN)

19:30 → 20:30 Dinner

Wednesday 8 October

07:45 → 08:45 Breakfast

08:45 → 09:45 Scientific computing on heterogeneous architectures

Programming for Heterogeneous Architectures - lecture 1

• Introduction to heterogeneous architectures and the performance challenge
• From general to specialized: Hardware accelerators and applications
• Type of workloads ideal for different accelerators
• Trade-offs between multi-core and many-core architectures
• Implications of heterogeneous hardware on the design and architecture of scientific software
• Embarrassingly parallel scientific applications in HPC and CERN

Speaker: Daniel Hugo Campora Perez (NVIDIA)

09:45 → 10:45 Optimizing existing large codebase - exercise 1

Optimisation of an existing, production grade large codebase

Speaker: Sebastien Ponce (CERN)

10:45 → 11:00 Announcements

11:00 → 11:15 Coffee break

11:15 → 12:15 Optimizing existing large codebase - exercise 2

Optimisation of an existing, production grade large codebase

Speaker: Sebastien Ponce (CERN)

12:15 → 13:00 Lunch

13:00 → 18:00 River rafting excursion - Cetina River

18:30 → 20:30 Dinner at Kastel Slanica Omis

20:30 → 21:30 Transport to Medils

Thursday 9 October

07:45 → 08:45 Breakfast

08:45 → 09:45 Performant programming for GPUs

Programming for Heterogeneous Architectures - lecture 2

• From SIMD to SPMD, a programming model transition
• Thread and memory organization
• Basic building blocks of a GPU program
• Control flow, synchronization, atomics

Speaker: Daniel Campora (NVIDIA)

09:45 → 10:45 Performant programming for GPUs

Programming for Heterogeneous Architectures - lecture 3

• Data locality, coalesced memory accesses, tiled data processing
• GPU streams, pipelined memory transfers
• Under the hood: branchless, warps, masked execution
• Debugging and profiling a GPU application

Speaker: Daniel Campora (NVIDIA)

10:45 → 11:00 Announcements

11:00 → 11:30 Coffee break

11:30 → 12:30 Practical vectorization

Parallel and Optimised Scientific Software - lecture 4

• Measuring vectorization level
• What to expect from vectorization
• Preparing code for vectorization
• Vectorizing techniques in C++: intrinsics, libraries, autovectorization

Speaker: Sebastien Ponce (CERN)

12:30 → 13:30 Practical vectorization - exercise

Speaker: Sebastien Ponce (CERN)

13:30 → 14:30 Lunch

14:30 → 16:30 Study time or sports

16:30 → 17:00 Coffee break

17:00 → 18:00 Programming for heterogeneous architectures - exercise

Speaker: Daniel Campora (NVIDIA)

18:00 → 19:00 Programming for heterogeneous architectures - exercise

Speaker: Daniel Campora (NVIDIA)

19:30 → 20:30 Dinner

Friday 10 October

07:45 → 08:45 Breakfast

08:45 → 09:45 Debrief of exercises

09:45 → 10:45 Design patterns and best practices

Programming for Heterogeneous Architectures - lecture 4

• Good practices: single precision, floating point rounding, avoid register spilling, prefer single source
• Other standards: SYCL, HIP, OpenCL
• Middleware libraries and cross-architecture compatibility
• Reusable parallel design patterns with real-life applications

Speaker: Daniel Campora (NVIDIA)

10:45 → 11:00 Announcements

11:00 → 11:30 Coffee break

11:30 → 12:30 Programming for heterogeneous architectures - exercise

Speaker: Daniel Campora (NVIDIA)

12:30 → 13:30 Programming for heterogeneous architectures - exercise

Speaker: Daniel Campora (NVIDIA)

13:30 → 14:30 Lunch

14:30 → 15:45 Exam

15:45 → 16:20 Coffee break

16:20 → 17:20 Closing Session

17:20 → 19:20 Free time

19:30 → 21:30 Closing dinner Kavanazona

Saturday 11 October

07:30 → 08:30 Breakfast

09:00 → 11:30 Departures